Thursday, January 23, 2020

Can BioNano Genomics Map A Way Past Twin Perils?

When playing with the structure of this piece in my mind, it occurred to me that Norman Maclean's thoughts about fly fishing apply just as well to biotechnology companies
It is also interesting that thoughts about fishing are often carried on in dialogue form where Hope and Fear -- or, many times, two Fears -- try to outweigh each other
The executive team at BioNano Genomics may well be gripped by this situation, as they are faced with two great perils: their finances and their markets.

BioNano has commercialized some technically impressive equipment. Their Saphyr optical mapping platform takes very large DNA molecules and entices them to thread themselves into nanochannels which hold them as elongated straight molecules.  It's like taking a bowl of rotini and tricking it into being dried spaghetti.  Then they perform nicking and nick translation to label tiny spots on these individual molecules and image that.  Computation then takes many such images and constructs maps from them.  It's hard not to find this all pretty amazing.  They've successfully taken that technology and shown it useful for assembling genomes as well as structural variant characterization in cancer and rare genetic disorders. On the business front, BioNano successfully executed an IPO in August of 2018, raising over $20 million - less than the original goal of $30 million but still not chump change.

Alas, that's where the good news pretty much ends.

The business front is the most immediately troubling.  Since the IPO in the $6 range the stock never passed $8 and has mostly followed a course that looks like the mountains around Mann Gulch, a steep slope downwards/  Currently hovering around $1.20, last fall it even broke below $1 for a number of days.

BioNano is listed on the NASDAQ, which has certain standards for membership -- one of which is to keep the share price above a dollar.  Another is to have a certain level of stockholder equity, and last summer the NASDAQ warned BioNano that they were in arrears of that standard and had six months to fix it.  Delisting would mean exile to the Pink Sheets, the dreaded neighborhood of thinly traded (and therefore easily manipulated) penny stocks and companies with dubious business plans and even more dubious management teams.  Not a place any respectable technology company ever wants to find themselves.
[2020-01-24 -- it's been pointed out that BioNano raised $18M in October, which I miserably failed to find in preparing this, and is now back in compliance with the capital requirements at NASDAQ as well as having that much more in their bank account]

The same third quarter SEC filing that mentions the delisting threat also shows a relatively grim balance sheet.  The end of September 2019 showed $8.2 million in the bank -- not much more than the $6.4 million loss recorded for that quarter.  There's $6.0 million owed them by customers (accounts receivable), which is greater than the $3.4 million owed their own creditors (accounts payable).  That roughly mirrors the quarterly revenue of $3.3 million being somewhat larger than the cost of revenue of $2.4 million.  But there's another $6.6 million in expenses -- R&D plus general overhead.  Which leads to a grim conclusion: if BioNano could eliminate overhead they'd be near breakeven, but with it they're nowhere close.

The science side is more of a medium term issue, not like the very immediate threats of delisting or running dry on cash, but certainly troublesome.  BioNano just got some good exposure at the Plant and Animal Genomes (PAG) conference and I see a couple of AGBT titles mentioning optical mapping.  That almost guarantees BioNano, as their competitor OpGen pivoted to an entirely different business.  Their only direct competitor would seem to be Nabsys 2.0, with their electronic mapping scheme.

But ultimately the BioNano Saphyr instrument is a very expensive benchtop unitasker.  If you have plenty of optical mapping needs, then you might be happy with it -- but in many labs it will see at best intermittent use.  So a niche instrument requiring unique training and maintenance.

Any mapping instrument faces threats from sequencing: sequencers are the ultimate multitaskers capable of a wide and growing array of tasks.  A unitasker that delivers data the generalists can't -- or maximizes some other value parameter such as cost, ease-of-workflow or wallclock time -- is kept, but little-by-little their position will erode.  At Warp we tried both OpGen and BioNano on our seemingly impossible bacterial assembly problems; OpGen bellyflopped and BioNano gave limited insight.  Then along came sufficiently long PacBio reads and our genomes snapped together.  Optical mapping just isn't something worth considering in the bacterial world -- though I'll note one of the AGBT talks is on optical mapping of virus genomes, which I'm curious to find out why that was seen as a good strategy.

Structural variants are a subject of intense interest in the sequencing world, so the experimentalists develop new protocols for DNA isolation to drive read lengths higher and the computationalists generate new algorithms to use that data to more precisely and sensitively detect SVs.  Several years ago some of the Nabsys folks showed me data suggesting that their platform was best for certain scales of SVs; in a world of 20kb HiFi reads and Oxford Nanopore libraries with read N50s in the 100 kilobase range I wonder at what scales that is still true.  10X Genomics just dropped out of the linked reads for genomes arena, but now there's Universal Sequencing Technology with their TELL-Seq kits.  So BioNano lost one competitor but has gained another.  Even if the clinical assays that BioNano and their collaborators have genotyped are small numbers of patients, the existence of those assays gives a map for competitors -- metrics to match or beat and data to compare against.

Another use of BioNano has been to help scaffold difficult large genomes.  In addition to the bottom-up threat of longer read lengths and better assemblies from long read and linked read systems, there's more of a top down problem of better-and-better HiC kits.  In their recent newsletter, Arima Genomics showed how the use of affinity capture can give resolution of 500 basepairs for HiC.  Since standard HiC kits don't even require initial DNA purification -- and hence can preserve very large structural features -- they have a chance to cover all the size ranges for structural variants not available to straight-up sequencing.  Oxford Nanopore has put forth their Pore-C scheme, which packs many proximity tags into a single read, enabling more pairwise interactions to be captured and opening up detecting N-way interactions.  A kit requiring no special equipment and no capital spent on unitasking equipment to clutter the bench would be a tough opponent for the mapping instrument companies; multiple such kits with open benchmarks is far worse.

Maclean had another thought that certainly covers BioNano
That's how you know when you have thought too much -- when you become a dialogue between You'll probably lose and You're sure to lose
BioNano is in desperate need of a major strategic alliance, even if they can't command great terms given their weakness (the probably lose arm).  They might stay independent, but their partner would need to inject a heap of capital in order to meet the NASDAQ capital minimum with lots of room for later share price fluctuation.  Far more likely is an outright acquisition by the likes of Thermo Fisher.  In the near term that would mean the sad but necessary laying off of much of BioNano's staff, with perhaps some limited R&D folks kept around to try to build kits for additional tissue types.  But the only way the financials would work in anyone's hands is to substantially trim that $6 million quarterly overhead cost so that the small gross profit on sales can actually turn into something of a net profit.  Any acquirer will also throw their marketing muscle at Saphyr to push more sales, though in a large conglomerate there is the risk that Saphyr would be just one more beast in a very chaotic product zoo.

There is one other curious asset that BioNano has -- a cross-license agreement with Pacific Biosciences for certain patents to use royalty-free in their own areas of expertise (mapping for BioNano, sequencing for PacBio) -- but also an option for a royalty-bearing license in the other company's field.  These agreements can be transferred as part of a change-in-control or merger.  So acquiring BioNano is a backdoor route to access certain PacBio patents, though getting patent rights is far different than getting all of PacBio's technology and skilled personnel.

The alternative to a strategic partnership is "sure to lose" -- bankruptcy and extinction as a corporate entity.  Well, that's perhaps a bit extreme -- with the right bits of luck and skill BioNano might be able to pull out of this dive.  As someone who's been part of biotechs whose coffers ran dry, I wish them well.  But they're going to need a lot more than well wishes to survive long term.


Anonymous said...

Novogene, The Mayo Clinic, UNC Chappell Hill, GeneDx, Perkins Elmer, University of Pennsylvania Hospital, Children’s Hospital of Philadelphia and Chicago, all don’t seem to have a problem with the “twin perils”. You have talked about all the things that other sequencing technologies can do similarly, but haven’t said anything about why it is that Bionano has that enables a moat to its IP assets. Also, you have given the critique of an entire company over a single SKU. Pretty narrow base to grow your opinion upon, considering that Saphyr is not the only product the company offers. It’s just one peace of the puzzle. Your opinion needs definite refinement.

Keith Robison said...

Thank you for your feedback. To address your three points
1) Happy customers are truly the votes BioNano needs. But customers also tend to be less focused on the business side -- one of the perils. Early adopters also tend to be the most committed. In any case, to overcome the business peril BioNano needs to attract more customers, and that's the hard part

2) If you have insight into a true IP moat that BioNano has that protects them from sequencing competitors, I'd like to hear it -- but realistically you cannot patent the concept of detecting SVs, only particular means.

3) The company has a single integrated product line around Saphyr -- look at their products page. There is Saphyr, consumables for Saphyr and software to analyze data generated by Saphyr -- and nothing else.

Anonymous said...

They did a capital raise of 18 million in Oct so they have more cash then you state. Also they just became compliant with nasdaq this week, you can check the Nasdaq noncompliance list and also confirmed by their investor relations liason.

Keith Robison said...

2nd commenter -- thank you, it is egregious I missed that fundraising of $20M, or about 3-4 quarters of funding at current loss rate. GenomeWeb article on October share sale

Anonymous said...

BioNano is having trouble "crossing the chasm". Interestingly, this usually happens pre-IPO and weeds out technology that doesn't have a market. If the SV-only market cannot support a technology that performs as it should (like BioNano), then why is Genturi spending all of their time and money to compete in the same market. Just because you can make cool technology does not mean you should. Genomic tools companies often make the same mistakes, they listen to their early-adopter minded tech leaders without an understanding of the market.

Anonymous said...

Unitasker? Curious. By your definition, a sequencer is a unitasker as well since it only sequences nucleic acids. Besides what's been published about Saphyr's ability to detect novel SVs, fusions, microdeletions, etc. genome wide, you are missing other applications of the technology...sizing large repeat expansions, identifying DNA replication firing events, providing methylation and structural information on the same molecules, elucidating fCNA with ecDNA in cancer...all published. This all on relatively new technology...use your imagination as to other unpublished work that may be happening by using their long unadulterated molecules.

Anonymous said...

It can never be as good as a long read sequencer. Long read sequencers get longer better cheaper data as time goes on. The capital cost is low.

Anonymous said...

I guess it's application specific. Show me a credible long read roadmap to getting at SVs at <5% allelic fraction across the whole genomes of cancers for <$500 per sample. Saphyr has published results on that showing clinical grade sensitivity/PPV. Sure, everyone would love one technology to rule all genomic applications, but despite over a decade of work and $billions spent, long read is great for some applications, but can't do what 60+ year old classical cyto methods can do. We need to be focused on getting better diagnostics and therapies to patients and stop being infatuated with any particular technology...IMHO.

Anonymous said...

Hi, thanks for the interesting post. Clarifying question: When you say that optical mapping is not suited for bacteria, why is that? Low complexity/lack of nicking sites? Cheers.

Antonio Pisani said...

BioNano finds itself on the cusp of major technology developments that will lead to an inflection point of its business.
Here a list of catalysts for 2020 :
1) DLE-2 and DLE-3 labeling enzymes with a 30℅ higher density than DLE-1
2) Sample multiplexing
3) 12-flowcell Saphyr chip
4) Allele Variant Fraction of 1℅
5):Specificity, Sensitivity and PPV targeted values of 100℅ for most SV's
6) targeted Crispr labeling
7) Loss of Heterozygosity
BioNano is also targeting the cytogenetics labs with the goal of establishing the Saphyr instrument as at least a complementary tool to the standard Fish aCGH and Southern Blotting.
Last but by no means least, BioNano's customers list includes the world's leading cancer company :Genentech (bought out by Roche for $46B).

Keith Robison said...

WRT bacteria, it's that good long read libraries will solve -- assemble to a closed chromosome -- the genomes except in very rare cases of really large duplications. I'd generally bet you can solve those with really careful nanopore library prep to get a library with 8X coverage of 100kb+ reads.

Anonymous said...


Maybe you would like to give an update on Bionano after reviewing the presentations from AGBT2020, as well as the new data-based business model.

You seemed to have missed the boat with your poorly-researched hit piece here.

Anonymous said...

the stock is now 60 cents. Seems author is right?

Thobias said...

It is always interesting to analyze stocks and what is going in the company. There are a lot of public companies to trade on the market. I think we can all agree that this is not a safe play. We can probably also agree that it might just be something really special. But to calculate the possibility for it to become huge is impossible. In a month they might have competition. In 3 month they might have concluded that it is not an efficient tool that brings something new or important. But they might just discover the opposite and become the most wanted tool which would bring them a solid status and possibilities to even develop this further in collaborations with major medical companies on the market. It all just becomes a guessing game and I am fine loosing my bet but I would not want to miss being in the stock at this time. (I have pretty much zero real knowledge about the business, just reading news and publications to make up my opinion on this one. So my respect to the ones with vast knowledge of the business. Sometimes though it is better to be somewhat uninformed in order to actually place som crazy bets:) Cheers!

Alex said...

my concern is that long sequencing can do sv. pacbio has put out a document showing 96% accuracy on some agricultural use case. bionano might be better for large human genome. but i can't understand if it is enough of a difference for anyone to care. if long read can do 85% on a human genome is that good enough? any thoughts?

Unknown said...

Long-read sequencing uses diced up samples which must then be reconstructed to produce a full genome. This "grass clippings" approach is a challenge when you consider the vast areas of the genome comprised of repeats. And for SV that occurs across various clippings containing repeats...well. You can begin to see that no matter how fast the processing becomes, traditional sequencers like PacBio will never be able to overcome these physics.

Long-read sequencing is preposterously short-read compared to Saphyr's read-length. And ever shall it be so.